LineChaser: A Smartphone-Based Navigation System for Blind People to Stand in Lines
Masaki Kuribayashi, Seita Kayukawa, Hironobu Takagi, Chieko Asakawa, Shigeo Morishima · 2021 · Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3411764.3445451
Summary
Standing in line is a mundane-looking task — at a cashier, bus stop, or airport check-in — but for a blind person it is a brittle, anxiety-ridden sequence: find where the line currently ends (a position that moves), join it without bumping into anyone, and then shuffle forward in sync with the person in front, all without the visual cue the rest of the queue relies on. Prior assistive navigation research has concentrated on reaching a fixed destination (a door, an exhibit, a landmark); line standing falls through the cracks because the 'destination' is the dynamically changing end of a human chain. This paper presents LineChaser, a smartphone-based system that runs on an off-the-shelf iPhone 11 Pro using only built-in sensors — RGB camera, infrared depth sensor, IMU, and ARKit-based visual inertial odometry localised against pre-placed AR markers. In Phase 1 the system guides the user to the entrance of a waiting area using an audio clock-position interface ('Walk to the 2 o'clock, 2.1 meters ahead'). In Phase 2 it runs YOLOv3-tiny at ~15 fps to detect the last person in line, locks onto that person using a Bhattacharyya-distance colour-histogram match to prevent target swaps, measures the user–target distance with the depth sensor, and vibrates the phone in distinct patterns to signal stop, move forward, and imminent collision. Notably, it defaults to a 1.7 m stop distance to preserve COVID-era social distancing. The paper reports a six-person preliminary study followed by a 12-person main user study across straight (L1) and serpentine (L2) lines with 2–4 sighted 'extras' moving intermittently.
Key findings
All 12 blind participants successfully found the end of the line in every one of 66 trials and successfully followed the moving line in every one of 72 trials — an end-to-end success that participants said was otherwise only achievable by asking a stranger. They stopped within the 1.5 m 'acceptable' zone 91.7% of the time while following the line (121/132) and 40.9% in the ideal centre position when first joining. Mean SUS score was 83.9 (grade A); the improved audio+vibration interface scored substantially higher than the vibration-only prototype (37.5, F). Confidence and comfort ratings rose significantly (Wilcoxon signed-rank, p<0.01) on every question about standing in a line in familiar and unfamiliar places, being confident not to bump into the person in front, and stress. Participants stopped just behind the target on 86.4% of line-movement trials in the preliminary study. Failures that did occur clustered into four causes: participants unable to hold the phone in the correct orientation, misunderstanding vibration signals, ignoring instructions, or ARKit localisation drift. Eleven of 12 participants were uncomfortable holding the phone pointed at strangers (anticipated social friction with the camera); P12, who had used the prototype, said LineChaser's handle and improved feedback made the phone manageable, a usability caveat rather than a rejection.
Relevance
For accessibility practitioners, this paper matters on two levels. First, it names an unsolved everyday task — line standing — that commercial navigation systems (Google Maps, Aira, BlindSquare, Soundscape) do not address and whose difficulty blind travellers routinely manage by asking strangers or skipping the task altogether. The finding that participants often avoid queuing independently ('I do not stand in lines by myself because it is troublesome') is a clear equity gap worth addressing. Second, the engineering pattern — off-the-shelf phone, AR-marker-based localisation, RGB+depth pedestrian tracking, dual audio/vibration feedback, built-in social-distancing default — is reproducible in operations where store staff can install a map and a few markers. Limitations: the deployment still requires pre-placed AR markers and a prepared floor map (same localisation dependency critiqued in static-route-map systems), it assumes an individual target rather than a small group standing together, and 11 of 12 participants flagged the social awkwardness of visibly pointing the phone's camera at the next person in line. Future wearable or rear-facing LiDAR (iPhone 12 Pro) form factors could mitigate that friction.
Tags: line detection · pedestrian detection · orientation and mobility · visual impairment · indoor navigation · smartphone assistive technology · social distancing · vibration feedback · blind navigation · wayfinding